mirror of
https://github.com/langbot-app/LangBot.git
synced 2026-06-13 09:16:04 +00:00
Feat/saas sandbox adaptation (#2234)
* fix(box): trust Box-reported skill paths when filesystem is not shared
In separated deployments (Docker Compose, k8s sidecar, --standalone-box,
remote runtime.endpoint) the Box runtime owns its own filesystem, so the
skill package_root it reports via list_skills is not resolvable on the
LangBot side. LangBot's reload_skills and build_skill_extra_mounts
validated those paths with os.path.isdir() against its own filesystem,
which silently dropped every skill in such deployments — breaking the
sandbox skill feature for the nsjail/SaaS backend.
Add BoxService.shares_filesystem_with_box, derived from the connector
transport (stdio = shared, WebSocket = separated), with an explicit
override seam for tests/embedders. Gate both isdir() guards on it: keep
local validation in shared-fs stdio mode, trust Box-reported paths
otherwise. The Box runtime only reports skills found on its own
filesystem, so those paths are valid there by construction.
Adds topology-derivation tests (real connector, no mocks) and
skill-retention tests for both shared and separated filesystems.
* build(docker): ship a self-contained nsjail sandbox backend in the image
Compile nsjail 3.6 from source in a dedicated multi-stage build and carry
only the binary plus its runtime libs (libprotobuf32, libnl-route-3-200)
into the final image. This lets the Box runtime isolate sandboxed code via
nsjail user/mount/pid/net namespaces without a host Docker socket — the
prerequisite for running Box on LangBot Cloud (k8s), where mounting
docker.sock would grant node root and is not acceptable for multi-tenant.
The build toolchain (build-essential/bison/flex/protobuf-dev/libnl-dev)
stays in the nsjail-build stage and is not present in the shipped image.
Verified: image builds (583MB), nsjail --help exits 0, libraries resolve,
and the real NsjailBackend executes an isolated command end-to-end on a
v6.1/cgroup2 host matching LangBot Cloud prod (rlimit fallback path, since
container /sys/fs/cgroup is read-only; PID-namespace isolation confirmed).
* feat(box): SaaS guard to force a single global sandbox scope
Add system.limitation.force_box_session_id_template: when non-empty it
overrides every pipeline's box-session-id-template at resolve time, pinning
all queries to one shared sandbox (e.g. {global}). This is the authoritative,
unbypassable guard — it runs on every exec call, so editing the pipeline
config via API cannot escape it. The web UI locks the Sandbox Scope selector
via a combined box_scope_editable flag (box available AND not forced).
* build(deps): pin langbot-plugin==0.4.2b1 (nsjail cgroup container-safety beta)
* fix(web): show forced sandbox scope + make disabled tooltip tap-friendly
When a SaaS deployment pins every pipeline to a fixed sandbox scope via
system.limitation.force_box_session_id_template, the Sandbox Scope selector was
correctly locked but still displayed the pipeline's stored value (e.g. the
per-chat default), misrepresenting the scope that the runtime actually enforces
on every exec. Coerce the displayed/saved value to the forced template so the
locked selector truthfully shows the active scope (e.g. Global).
Also fix the disabled_tooltip being invisible on touch devices: hover-only Radix
tooltips never open without a pointer, so the explanation of why the field is
locked could not be read on mobile. Wrap the info icon so a tap toggles the
tooltip while desktop hover still works.
* feat(web): hide sidebar new-version prompt for edition=cloud
Cloud instances are upgraded centrally by the operator, so surfacing a GitHub
'new version available' badge to tenants is misleading and actionable only by
the operator. Skip the release check entirely when edition=cloud.
* style(web): prettier formatting for DisabledTooltipIcon ternary
* chore(deps): bump langbot-plugin to 0.4.2b2
Picks up the SDK fix that creates a read-write host_path before the
nsjail bind-mount, fixing the SaaS MCP shared-workspace sandbox failure
(exec exit 255 with empty output when host_path didn't exist).
* chore(deps): bump langbot-plugin to 0.4.2b3
Picks up the nsjail /dev-node fix so stdio MCP servers (uvx-launched) can
start under force_global_sandbox instead of failing with 'Connection closed
/ please check URL'.
* fix(web): show real MCP runtime status on installed extensions list
The installed-extensions list badge keyed solely off the enable flag, so a
server that was still CONNECTING (or in ERROR) was shown as 'Connected'.
Reflect the actual runtime_info.status (connecting/connected/error/disabled)
with matching colors, and poll quietly every 3s while any MCP server is
connecting so the badge transitions without a manual refresh.
* chore(deps): bump langbot-plugin to 0.4.2b4
Picks up the 30s start_managed_process timeout so cold uvx MCP bootstraps
don't get torn down mid-install.
* style(web): satisfy prettier — parenthesize nullish-coalescing in ternary
* fix(mcp): isolate transient test sessions from the shared Box session
A config-page 'test' (server_name='_', no persisted UUID) ran in the same
shared 'mcp-shared' Box session as live MCP servers. A failing test (e.g.
empty args) churned that shared session and tore down healthy, already-
connected servers — leaving them stuck after exhausting their retries.
Mark UUID-less sessions as transient, give them their own isolated Box
session ('mcp-test-<uuid>'), and fully delete that session on cleanup so
tests can never disturb live servers and don't leak sessions.
* fix(mcp): tear down transient test session after test completes
A successful config-page test left its isolated 'mcp-test-<uuid>' Box
session running (the lifecycle task blocks until shutdown). Wrap the
transient test coroutine so it always shuts the session down afterward,
preventing isolated test sessions from leaking.
This commit is contained in:
Junyan Chin
@@ -198,6 +198,35 @@ function WebhookUrlField({
|
||||
);
|
||||
}
|
||||
|
||||
// Hover-only Radix tooltips never open on touch devices (no pointer hover),
|
||||
// so the ``disabled_tooltip`` explaining why a field is locked was invisible on
|
||||
// mobile. This wrapper makes the info icon also toggle the tooltip on tap while
|
||||
// keeping hover behavior on desktop.
|
||||
function DisabledTooltipIcon({ text }: { text: string }) {
|
||||
const [open, setOpen] = useState(false);
|
||||
return (
|
||||
<TooltipProvider delayDuration={100}>
|
||||
<Tooltip open={open} onOpenChange={setOpen}>
|
||||
<TooltipTrigger asChild>
|
||||
<button
|
||||
type="button"
|
||||
aria-label={text}
|
||||
className="inline-flex shrink-0"
|
||||
onClick={(e) => {
|
||||
e.preventDefault();
|
||||
e.stopPropagation();
|
||||
setOpen((v) => !v);
|
||||
}}
|
||||
>
|
||||
<Info className="h-3.5 w-3.5 text-muted-foreground cursor-help shrink-0" />
|
||||
</button>
|
||||
</TooltipTrigger>
|
||||
<TooltipContent className="max-w-xs">{text}</TooltipContent>
|
||||
</Tooltip>
|
||||
</TooltipProvider>
|
||||
);
|
||||
}
|
||||
|
||||
export default function DynamicFormComponent({
|
||||
itemConfigList,
|
||||
onSubmit,
|
||||
@@ -551,16 +580,7 @@ export default function DynamicFormComponent({
|
||||
: '';
|
||||
const renderDisabledTooltipIcon = () =>
|
||||
disabledTooltip ? (
|
||||
<TooltipProvider delayDuration={100}>
|
||||
<Tooltip>
|
||||
<TooltipTrigger asChild>
|
||||
<Info className="h-3.5 w-3.5 text-muted-foreground cursor-help shrink-0" />
|
||||
</TooltipTrigger>
|
||||
<TooltipContent className="max-w-xs">
|
||||
{disabledTooltip}
|
||||
</TooltipContent>
|
||||
</Tooltip>
|
||||
</TooltipProvider>
|
||||
<DisabledTooltipIcon text={disabledTooltip} />
|
||||
) : null;
|
||||
|
||||
// Webhook URL fields are display-only; render outside of form binding
|
||||
|
||||
@@ -1674,24 +1674,31 @@ export default function HomeSidebar({
|
||||
.catch(() => {});
|
||||
}
|
||||
|
||||
getCloudServiceClientSync()
|
||||
.getLangBotReleases()
|
||||
.then((releases) => {
|
||||
if (releases && releases.length > 0) {
|
||||
const latestStable = releases.find((r) => !r.prerelease && !r.draft);
|
||||
const latest = latestStable || releases[0];
|
||||
setLatestRelease(latest);
|
||||
// Cloud edition is updated centrally by the operator, so end users should
|
||||
// not see a "new version available" prompt in the sidebar. Skip the GitHub
|
||||
// release check entirely for edition=cloud.
|
||||
if (systemInfo?.edition !== 'cloud') {
|
||||
getCloudServiceClientSync()
|
||||
.getLangBotReleases()
|
||||
.then((releases) => {
|
||||
if (releases && releases.length > 0) {
|
||||
const latestStable = releases.find(
|
||||
(r) => !r.prerelease && !r.draft,
|
||||
);
|
||||
const latest = latestStable || releases[0];
|
||||
setLatestRelease(latest);
|
||||
|
||||
const currentVersion = systemInfo?.version;
|
||||
if (currentVersion && latest.tag_name) {
|
||||
const isNewer = compareVersions(latest.tag_name, currentVersion);
|
||||
setHasNewVersion(isNewer);
|
||||
const currentVersion = systemInfo?.version;
|
||||
if (currentVersion && latest.tag_name) {
|
||||
const isNewer = compareVersions(latest.tag_name, currentVersion);
|
||||
setHasNewVersion(isNewer);
|
||||
}
|
||||
}
|
||||
}
|
||||
})
|
||||
.catch((error) => {
|
||||
console.error('Failed to fetch releases:', error);
|
||||
});
|
||||
})
|
||||
.catch((error) => {
|
||||
console.error('Failed to fetch releases:', error);
|
||||
});
|
||||
}
|
||||
|
||||
getCloudServiceClientSync()
|
||||
.getGitHubRepoInfo()
|
||||
|
||||
@@ -8,6 +8,7 @@ import {
|
||||
import DynamicFormComponent from '@/app/home/components/dynamic-form/DynamicFormComponent';
|
||||
import N8nAuthFormComponent from '@/app/home/components/dynamic-form/N8nAuthFormComponent';
|
||||
import { useBoxStatus } from '@/app/infra/hooks/useBoxStatus';
|
||||
import { systemInfo } from '@/app/infra/http';
|
||||
import { Button } from '@/components/ui/button';
|
||||
import { useForm } from 'react-hook-form';
|
||||
import { zodResolver } from '@hookform/resolvers/zod';
|
||||
@@ -420,11 +421,41 @@ export default function PipelineFormComponent({
|
||||
// opt-in via ``disable_if`` + ``disabled_tooltip`` rather than every page
|
||||
// hard-coding a banner. Field-level gating keeps unrelated fields
|
||||
// untouched.
|
||||
//
|
||||
// ``box_scope_editable`` folds the two reasons the Sandbox Scope selector
|
||||
// can be locked into a single flag the yaml ``disable_if`` consumes:
|
||||
// 1. Box sandbox is unavailable, or
|
||||
// 2. the deployment pins all pipelines to a fixed scope via
|
||||
// ``system.limitation.force_box_session_id_template`` (SaaS).
|
||||
const forcedBoxTemplate =
|
||||
systemInfo.limitation?.force_box_session_id_template || '';
|
||||
const boxScopeForced = !!forcedBoxTemplate;
|
||||
const stageSystemContext =
|
||||
stage.name === 'local-agent'
|
||||
? { box_available: boxAvailable }
|
||||
? {
|
||||
box_available: boxAvailable,
|
||||
box_scope_editable: boxAvailable && !boxScopeForced,
|
||||
}
|
||||
: undefined;
|
||||
|
||||
// When the deployment pins every pipeline to a fixed sandbox scope (SaaS
|
||||
// ``force_box_session_id_template``), the Sandbox Scope selector is locked.
|
||||
// The runtime already overrides the scope on every exec, but the stored
|
||||
// pipeline value can be anything (e.g. the per-chat default), which would
|
||||
// make the locked selector display a scope that is NOT the one actually in
|
||||
// effect. Coerce the displayed/saved value to the forced template so the UI
|
||||
// truthfully reflects runtime behavior.
|
||||
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
||||
const stageInitialValues: Record<string, any> =
|
||||
(form.watch(formName) as Record<string, any>)?.[stage.name] || {};
|
||||
const effectiveInitialValues =
|
||||
stage.name === 'local-agent' && boxScopeForced
|
||||
? {
|
||||
...stageInitialValues,
|
||||
'box-session-id-template': forcedBoxTemplate,
|
||||
}
|
||||
: stageInitialValues;
|
||||
|
||||
return (
|
||||
<Card key={stage.name}>
|
||||
<CardHeader>
|
||||
@@ -438,10 +469,7 @@ export default function PipelineFormComponent({
|
||||
<CardContent className="space-y-6">
|
||||
<DynamicFormComponent
|
||||
itemConfigList={stage.config}
|
||||
initialValues={
|
||||
// eslint-disable-next-line @typescript-eslint/no-explicit-any
|
||||
(form.watch(formName) as Record<string, any>)?.[stage.name] || {}
|
||||
}
|
||||
initialValues={effectiveInitialValues}
|
||||
onSubmit={(values) => {
|
||||
handleDynamicFormEmit(formName, stage.name, values);
|
||||
}}
|
||||
|
||||
@@ -160,16 +160,34 @@ export default function ExtensionCardComponent({
|
||||
{cardVO.mode.toUpperCase()}
|
||||
</Badge>
|
||||
)}
|
||||
<Badge
|
||||
variant="outline"
|
||||
className={`text-[0.7rem] flex-shrink-0 ${
|
||||
cardVO.enabled
|
||||
? 'border-green-400 text-green-600 dark:text-green-400'
|
||||
: 'border-gray-400 text-gray-600 dark:text-gray-300'
|
||||
}`}
|
||||
>
|
||||
{cardVO.enabled ? t('mcp.statusConnected') : t('mcp.statusDisabled')}
|
||||
</Badge>
|
||||
{(() => {
|
||||
// Reflect the real runtime status, not just the enabled flag.
|
||||
// A server can be enabled but still CONNECTING or in ERROR — showing
|
||||
// "Connected" in those cases is misleading.
|
||||
const runtime = cardVO.enabled
|
||||
? (cardVO.runtimeStatus ?? 'connecting')
|
||||
: 'disabled';
|
||||
const badgeClass: Record<string, string> = {
|
||||
connected: 'border-green-400 text-green-600 dark:text-green-400',
|
||||
connecting: 'border-amber-400 text-amber-600 dark:text-amber-400',
|
||||
error: 'border-red-400 text-red-600 dark:text-red-400',
|
||||
disabled: 'border-gray-400 text-gray-600 dark:text-gray-300',
|
||||
};
|
||||
const badgeLabel: Record<string, string> = {
|
||||
connected: t('mcp.statusConnected'),
|
||||
connecting: t('mcp.connecting'),
|
||||
error: t('mcp.statusError'),
|
||||
disabled: t('mcp.statusDisabled'),
|
||||
};
|
||||
return (
|
||||
<Badge
|
||||
variant="outline"
|
||||
className={`text-[0.7rem] flex-shrink-0 ${badgeClass[runtime] ?? badgeClass.disabled}`}
|
||||
>
|
||||
{badgeLabel[runtime] ?? badgeLabel.disabled}
|
||||
</Badge>
|
||||
);
|
||||
})()}
|
||||
</div>
|
||||
<div className="text-[0.8rem] text-muted-foreground line-clamp-2 w-full">
|
||||
{cardVO.description ||
|
||||
|
||||
@@ -103,8 +103,8 @@ const PluginInstalledComponent = forwardRef<
|
||||
getExtensionList();
|
||||
}
|
||||
|
||||
async function getExtensionList() {
|
||||
setLoading(true);
|
||||
async function getExtensionList(silent = false) {
|
||||
if (!silent) setLoading(true);
|
||||
try {
|
||||
const client = getCloudServiceClientSync();
|
||||
|
||||
@@ -200,12 +200,25 @@ const PluginInstalledComponent = forwardRef<
|
||||
setExtensionList(extensions);
|
||||
} catch (error) {
|
||||
console.error('Failed to fetch extension list:', error);
|
||||
setExtensionList([]);
|
||||
if (!silent) setExtensionList([]);
|
||||
} finally {
|
||||
setLoading(false);
|
||||
if (!silent) setLoading(false);
|
||||
}
|
||||
}
|
||||
|
||||
// While any MCP server is still connecting, poll quietly so the status badge
|
||||
// transitions (connecting -> connected/error) without a manual refresh.
|
||||
useEffect(() => {
|
||||
const hasConnecting = extensionList.some(
|
||||
(e) => e.type === 'mcp' && e.enabled && e.runtimeStatus === 'connecting',
|
||||
);
|
||||
if (!hasConnecting) return;
|
||||
const timer = setInterval(() => {
|
||||
getExtensionList(true);
|
||||
}, 3000);
|
||||
return () => clearInterval(timer);
|
||||
}, [extensionList]);
|
||||
|
||||
useImperativeHandle(ref, () => ({
|
||||
refreshPluginList: getExtensionList,
|
||||
}));
|
||||
|
||||
@@ -325,6 +325,10 @@ export interface SystemLimitation {
|
||||
max_bots: number;
|
||||
max_pipelines: number;
|
||||
max_extensions: number;
|
||||
/** When non-empty, every pipeline is forced to this Box sandbox-scope
|
||||
* template (e.g. ``{global}``) and the per-pipeline "Sandbox Scope"
|
||||
* selector is locked. Used by SaaS deployments. Empty = no restriction. */
|
||||
force_box_session_id_template?: string;
|
||||
}
|
||||
|
||||
export interface WizardProgress {
|
||||
|
||||
Reference in New Issue
Block a user